Nvidia (Nasdaq: NVDA) jumps after $96bn Q2 as $108bn Q3 test looms

NVIDIA Corporation (Nasdaq: NVDA) has delivered another extraordinary acceleration in artificial intelligence infrastructure demand, reporting fiscal second-quarter revenue of US$96.2 billion, up 106% year on year, and guiding to approximately US$108 billion for Q3. Data Center revenue reached US$89.0 billion, up 117%, while GAAP operating income more than doubled to US$63.7 billion and net income reached US$59.7 billion. NVDA closed the August 26 regular session at US$209.89 before the results, down about 1.5%, but subsequently rose roughly 4% in extended trading as investors absorbed the earnings beat, stronger guidance and management’s expectation that AI demand can remain exceptionally strong into fiscal 2028. The next investment test is no longer whether Nvidia can grow rapidly, but whether a company worth more than US$5 trillion can sustain extraordinary revenue growth while gross margins begin retreating from their recent peak.

Why did Nvidia shares recover after initially struggling with the Q2 results?

Nvidia entered the report carrying unusually high expectations. The shares had already fallen from US$225.16 on August 14 to US$209.89 on August 26 as investors debated whether even another large earnings beat would be sufficient to restart the AI trade.

The numbers cleared that hurdle comfortably. Q2 revenue increased 18% sequentially from US$81.6 billion to US$96.2 billion and more than doubled from US$46.7 billion a year earlier. Wall Street had been expecting revenue around US$92 billion.

Non-GAAP diluted earnings per share reached US$2.22 compared with approximately US$2.09 expected by analysts. GAAP diluted EPS increased 128% year on year to US$2.46.

Operating leverage remained exceptional. GAAP operating income increased 124% to US$63.7 billion, while non-GAAP operating income reached US$64.0 billion. Nvidia generated an operating profit equal to roughly two-thirds of quarterly revenue.

The initial after-hours response was mixed because investors focused on the next stage of margin pressure and the extremely high expectations already embedded in the stock. The reaction strengthened during the earnings call as management outlined continued supply constraints, accelerating adoption of the Vera Rubin platform and an unusually bullish longer-term revenue outlook.

NVDA ultimately traded around US$218 in extended dealing, roughly 4% above the regular-session close.

The wider price context remains less euphoric. The US$209.89 regular close was approximately 3.2% below the August 20 close of US$216.85, although it remained about 6.8% above the July 27 close of US$196.51. The stock’s current 52-week range is approximately US$164.07 to US$236.54.

That places Nvidia roughly 11% below its 52-week high even after delivering one of the strongest quarterly growth rates among the world’s largest companies.

What does Nvidia’s US$108bn Q3 guidance imply for the AI growth cycle?

Nvidia expects Q3 revenue of US$108 billion, plus or minus 2%.

At the midpoint, that represents approximately 12.2% sequential growth from Q2. If achieved, revenue for the first nine months of fiscal 2027 would reach approximately US$285.8 billion.

For perspective, Nvidia generated US$215.9 billion during the entirety of fiscal 2026.

The company would therefore surpass last year’s full-year revenue by roughly US$70 billion before even entering the final quarter of fiscal 2027.

Another notable feature of the guidance is what it excludes. Nvidia has assumed no Data Center compute revenue from China in its US$108 billion forecast.

That means the guidance does not depend on a recovery in Chinese accelerator sales following years of changing U.S. export restrictions. If Nvidia eventually receives approval to sell additional products into China and customers are willing to purchase them, that could represent incremental revenue rather than revenue already embedded in the Q3 target.

The underlying demand picture remains heavily concentrated around AI infrastructure.

Data Center revenue of US$89 billion represented roughly 92.5% of total Q2 revenue. The segment increased 18% sequentially and 117% year on year.

That concentration has produced Nvidia’s extraordinary financial growth, but it also means investors are effectively making a very large bet on the duration of global AI infrastructure spending.

Management’s confidence extends beyond the current fiscal year. During the earnings call, Nvidia projected approximately 70% revenue growth for the fiscal year ending in January 2028, substantially above the growth rate many investors previously assumed would follow the current expansion.

If that outlook proves broadly correct, the current AI investment cycle would be extending much further than the conventional semiconductor booms that historically lasted only a few years.

Can Vera Rubin extend Nvidia’s dominance beyond the Blackwell cycle?

Vera Rubin is rapidly becoming the most important product transition investors need to monitor.

Nvidia said the platform is now in full production, with systems running at partners including Microsoft Azure, Google Cloud, Oracle Cloud Infrastructure, CoreWeave and Nebius.

Management indicated during the earnings discussion that Rubin could represent approximately 20% of Data Center revenue during Q3 as production ramps.

Applying 20% mechanically to Q2’s US$89 billion Data Center revenue would already represent an enormous revenue base, although the actual Q3 segment total will differ and management has not provided a precise Rubin dollar forecast.

The importance of the transition is strategic rather than merely numerical.

One of the largest risks to high-growth semiconductor companies is a pause when customers wait for the next architecture. Nvidia is attempting to avoid that dynamic by moving rapidly from Blackwell into Vera Rubin while continuing to expand networking, CPUs, software and complete AI-factory systems.

The company is also pushing beyond individual GPUs.

Nvidia announced Spectrum-6 networking systems, the Vera CPU, the Groq 3 LPX inference accelerator, BlueField infrastructure products and the DSX platform for designing and operating large AI factories.

That breadth creates a larger addressable market per data centre. Instead of selling only the accelerator, Nvidia increasingly supplies compute, networking, CPUs, software and infrastructure architecture.

It also makes Nvidia more strategically important to customers, while increasing the economic incentive for competitors and hyperscalers to develop alternatives.

Amazon, Google, Microsoft and other large customers continue designing custom AI processors. AMD is also expanding its accelerator portfolio. Nvidia therefore needs Rubin to deliver substantial performance and cost advantages rather than relying on the installed CUDA ecosystem alone.

Why could Nvidia’s falling gross margin become the next major valuation issue?

The strongest cautionary signal in the Q2 results is not revenue growth. It is the direction of gross margin.

GAAP and non-GAAP gross margins were both 75.0% during Q2. Nvidia expects approximately 74.0%, plus or minus 50 basis points, during Q3.

Management has also warned that rapidly increasing memory and component costs could push margins lower again during Q4 before they begin recovering.

Commentary following the results indicated a potential trough around 71% to 72%.

A move from 75% to 72% may appear relatively small when revenue is still growing rapidly, but three percentage points applied to a US$100 billion-plus quarterly revenue base represents billions of dollars of gross profit.

The pressure partly reflects Nvidia’s own success.

AI servers require enormous quantities of advanced memory, networking and supporting hardware. As demand for those components rises across the industry, Nvidia faces higher procurement costs even while it maintains exceptional pricing power over the overall platform.

That creates a different earnings setup from the past several quarters.

Until now, revenue growth and margin expansion often reinforced one another. The next stage could involve extraordinary top-line growth occurring alongside some margin compression.

Investors should therefore avoid treating revenue growth alone as the measure of execution.

If Nvidia can grow Q3 revenue toward US$108 billion while maintaining gross margins around 74%, earnings should remain exceptionally strong. If component inflation pushes margins substantially below management’s expected range, profit growth could begin trailing revenue growth by a wider margin.

Is Nvidia expensive above a US$5tn market valuation?

At the August 26 regular close of approximately US$209.89, Nvidia carried an equity market capitalisation of roughly US$5.08 trillion.

That scale is unprecedented for a semiconductor company and leaves very little room for investors to think about Nvidia as an overlooked growth stock.

Yet the earnings multiple is less extreme than the market capitalisation alone might suggest.

Nvidia generated GAAP diluted earnings per share of US$1.30 in Q3 FY26, US$1.76 in Q4 FY26, US$2.39 in Q1 FY27 and US$2.46 in Q2 FY27. Adding those four reported quarters produces mechanically calculated trailing GAAP EPS of approximately US$7.91.

At US$209.89, that equates to roughly 26.5 times trailing GAAP earnings.

Using an extended-trading price around US$218 raises the multiple to approximately 27.6 times.

Those ratios are not cheap in isolation, but they are unusual for a company whose latest quarterly revenue increased 106% and whose next-quarter guidance implies another double-digit sequential increase.

The more difficult valuation question is therefore not whether 26 to 28 times earnings looks excessive relative to current growth. It is what earnings multiple Nvidia deserves after growth eventually normalises.

If revenue can still increase around 70% in fiscal 2028, the current valuation could compress rapidly through earnings growth even without a large share-price increase.

If AI infrastructure spending slows much sooner, the opposite happens. Nvidia’s absolute market value means that even a reasonable earnings multiple can produce enormous changes in shareholder value when long-term growth assumptions move.

This makes the duration of AI spending more important than conventional quarter-to-quarter valuation comparisons.

Does Nvidia have enough cash to support AI infrastructure and shareholder returns?

Nvidia’s balance sheet remains exceptionally strong despite the increasing capital requirements associated with the AI ecosystem.

The company ended Q2 with US$22.4 billion of cash and equivalents and US$34.1 billion of marketable debt securities. It also held approximately US$42.8 billion of marketable equity securities.

Short-term and long-term debt totalled approximately US$33.4 billion.

More importantly, the underlying business continues producing enormous cash flows. Free cash flow reached US$21.3 billion during Q2 and US$69.9 billion across the first six months of fiscal 2027.

Nvidia returned approximately US$26 billion to shareholders during Q2 through share repurchases and dividends and still had around US$99 billion remaining under its authorised buyback programme.

The company has also increased its quarterly dividend to US$0.25 per share, payable October 1 to shareholders of record on September 10.

At the same time, Nvidia is becoming more directly involved in enabling AI infrastructure investment.

It has announced partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR intended to mobilise more than US$500 billion of third-party capital for AI infrastructure over time.

That strategy creates an important new consideration.

Helping customers and infrastructure partners finance AI capacity can accelerate demand for Nvidia systems, but investors need to distinguish ordinary product demand from situations where Nvidia is increasingly supporting the financing ecosystem that ultimately purchases its technology.

Nvidia’s financial exposure remains modest relative to its cash generation and market value today. The issue becomes more important if such arrangements grow materially over time.

What could derail Nvidia’s US$108bn Q3 growth trajectory?

The largest risk remains a slowdown in AI capital expenditure.

Nvidia’s customers include some of the world’s largest and financially strongest technology companies, but those businesses still need to demonstrate economic returns from hundreds of billions of dollars of AI infrastructure investment.

If AI services generate enough revenue and productivity gains, spending could continue expanding for years. If utilisation or monetisation falls short, capital budgets could eventually be reconsidered.

The second risk is margin compression.

Memory prices and other component costs are rising while Nvidia moves into increasingly complex complete systems. Revenue can continue beating expectations while profit expectations disappoint if margins fall faster than anticipated.

The third risk is competition and customer concentration. Nvidia’s largest customers have the resources and strategic motivation to develop their own accelerators, while AMD and other semiconductor companies are targeting the same opportunity.

China remains another uncertainty, although its immediate impact is reduced because Nvidia has assumed zero China Data Center compute revenue in the Q3 outlook.

None of those risks currently appears to be slowing overall demand materially. The question is whether that remains true as Nvidia’s quarterly revenue moves beyond US$100 billion and the industry’s capital requirements become progressively larger.

Nvidia stock key takeaways after Q2 FY27 earnings

• Nvidia reported Q2 FY27 revenue of US$96.2 billion, up 106% year on year and 18% sequentially, while Data Center revenue increased 117% to US$89.0 billion.
• GAAP operating income reached US$63.7 billion and net income increased 126% to US$59.7 billion, with diluted GAAP EPS of US$2.46.
• Q3 revenue guidance of US$108 billion implies approximately 12.2% sequential growth and excludes any Data Center compute revenue from China.
• Nvidia’s regular-session August 26 close of US$209.89 was about 3.2% below its August 20 close but approximately 6.8% above July 27, before the stock gained roughly 4% in extended trading after earnings.
• Vera Rubin is already in full production and is expected to become a meaningful part of Data Center sales during Q3, creating the next major product-transition test.
• Gross margin is expected to decline from 75% in Q2 to approximately 74% in Q3, with higher memory and component costs potentially creating additional pressure during Q4.
• At the regular-session close, Nvidia was valued around US$5.08 trillion and at roughly 26.5 times mechanically calculated trailing GAAP EPS of US$7.91.

What would strengthen or weaken the Nvidia investment case from here?

Nvidia has once again produced financial growth that would normally be associated with a much smaller company. Revenue doubled, Data Center sales increased 117%, operating income more than doubled and management’s US$108 billion Q3 forecast implies that the company has still not reached a near-term revenue plateau.

The investment case would strengthen if Q3 revenue reaches or exceeds the US$108 billion midpoint, Vera Rubin ramps without disrupting Blackwell demand and gross margin remains close to the guided 74% level. Continued evidence that AI labs, cloud providers, enterprises and sovereign customers are all expanding infrastructure simultaneously would support management’s argument that the AI buildout remains early rather than mature.

The thesis would weaken if higher memory costs push margins below guidance, major hyperscalers begin slowing capital expenditure or Rubin demand reflects replacement of existing Nvidia systems rather than expansion of total infrastructure. A widening dependence on financing arrangements designed to support AI-factory construction would also deserve closer scrutiny if Nvidia’s own financial exposure begins rising materially.

The most important change after Q2 is that Nvidia has raised the scale of the debate again. The company is no longer trying to prove it can generate US$50 billion or even US$80 billion of quarterly revenue. It is guiding toward US$108 billion while suggesting another year of exceptional growth beyond the current fiscal period.

At more than US$5 trillion of market value, the stock does not require merely another earnings beat. It requires evidence that the AI infrastructure cycle can remain powerful long enough for today’s extraordinary revenue base to become the foundation for another leg of growth rather than the peak against which future quarters are measured.